Determining in-situ stress profiles of hydrocarbon reservoirs from geophysical well logs using intelligent systems

This work presents a new and novel technique for determining the in-situ stress profile of hydrocarbon reservoirs from geophysical well logs using a combination of fuzzy logic and neural networks. It is well established, that in-situ stress cannot be generated from well logs alone. This is because two sets of formations may have very similar geologic signatures but possess different in-situ stress profiles because of varying degrees of tectonic activities in each region. By using two new parameters as surrogates for tectonic activities, fuzzy logic to interpret the logs and rank parameter influence, and neural network as a mapping tool, it has become possible to accurately generate in-situ stress profiles. This paper demonstrates the superiority of this new approach over conventional approaches used in the oil and gas industry.